سامانه همایش ها

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

ECG-Based Detection of Acute Myocardial Infarction Using a Wrist-Worn Device: a Machine Learning Approach

نویسندگان :

Tania Hossein Khani¹ Amir hossein Tajarrod² Asghar Zarei³ Mousa Shamsi⁴

1- دانشکده مهندسی پزشکی، دانشگاه صنعتی سهند، تبریز، ایران 2- دانشکده مهندسی پزشکی، دانشگاه صنعتی سهند، تبریز، ایران 3- دانشکده مهندسی پزشکی، دانشگاه صنعتی سهند، تبریز، ایران 4- دانشکده مهندسی پزشکی، دانشگاه صنعتی سهند، تبریز، ایران

کلمات کلیدی :

Machine learning،Acute myocardial infarction،ECG،Wrist-worn wearable ECG،Hjorth parameters

چکیده :

Identifying acute myocardial infarction (AMI) at an early stage, particularly outside the hospital, remains one of the most pressing challenges in modern healthcare. While many wearable devices can record electrocardiogram (ECG) signals, most lack the essential precordial leads that are critical for accurate AMI detection. In this study, we evaluate the diagnostic capability of a wrist-worn, two-lead wearable ECG (wECG) device and compare its performance with the clinical standard, the conventional 12-lead ECG. Our analysis is based on a dataset where wECG and standard 12-lead ECG signals were recorded simultaneously from three participant groups: healthy individuals (CTRL), patients diagnosed with AMI, and patients with other cardiovascular diseases (CVD). This paper proposes a framework for diagnosing AMI patients as distinct from healthy individuals. Within this framework, we extracted both statistical features and Hjorth parameters. Then employed four different machine learning classifiers to assess classification performance across various scenarios. Using mutual information and f-test scores, we selected the best lead based on inter-class separation. The standard 12-lead ECG models achieved nearly flawless results, reaching 100% average accuracy. The wECG device also demonstrated impressive capabilities, accurately distinguishing between healthy participants and AMI patients with more than 98% average accuracy. Notably, the V5-LA configuration, when processed with the KNN classifier, achieved perfect average accuracy, highlighting the strong diagnostic power of this single lead. Overall, our results indicate that with careful design, a compact wECG device has the potential to serve as a reliable and highly effective tool for AMI detection in pre-hospital environments.